[Top] [All Lists]

Re: [Amps] Grid Dipping the Pi network in a new amp?

To: "'STEVEN & NANCY FRAASCH'" <>,"'Robert Bonner'" <>, "'Tom W8JI'" <>
Subject: Re: [Amps] Grid Dipping the Pi network in a new amp?
From: "Chuck Curran" <>
Date: Tue, 17 Jul 2007 21:22:56 -0500
List-post: <>
To All:

I have received quite a bit of welcomed advice, again, thanks to all who
have taken the time to make suggestions and recommendations.

A little more information was requested on this amp design, so I had better
get to it.  Just to repeat for any new readers, I tried grid dipping the
tank circuit and it just won't dip.  

The amps is based on a pair of 3-400Z's, which I have owned since new in
1969 -- they ran in a 10 meter linear for one DX CW contest back in 1970, or
about 48 hours.  Then back in the box.

All of the questions, I hope to answer with this description.  I have a
brand new National R-175A, as the plate choke, feeding into a pair of 500 pf
doorknob caps.  A 300 pf vacuum variable tune cap into the B&W 852 tank
coil, then a 1300 pf vacuum variable with a padding cap to get up to 1800 pf
on 80 meters.  I have checked the tank coil using two impedance bridges,
getting 6.9 and 7.1 uH against a B&W spec of 7.0 uH for the 80 meter
inductance.  All was checked for shorts to ground.  Each cap was also
checked,  and they can be varied from minimum to maximum capacitance without
any shorts or other negative items being detected.  

I will have 3,150 VDC no load, coming from a 5 Kw distribution transformer
that was unwound down to 2150 VAC  -- lots of fun. (That's another story)  I
am figuring about 2900 VDC under load, at 800 ma.  So the resulting plate
impedance, using the formula Rp = Ep/2*Ip gives around 1800 ohms.  The
specific B&W 852 tank coil was selected, since it was designed to work with
a low voltage high current system, which I believe I have.  Do the
calculations yourself and you will find that this results in a design Q of
12, the good ol' magic number.  I fully agree with the observations made by
one of the poster, this tank coil would be bad news with a 4500 VDC plate
voltage and low current.  However, I feel it has a high probability of
working , within it's intended work envelope.  We will see soon, I hope!

I have little time, the amp has been on the bench for more than 10 months,
but this Saturday I intend to pull out the new MFJ-259B antenna analyzer and
see what I can determine.  First, I will make darn sure that all of the
brass strap connecting the vacuum variables is making a proper connection,
without any shorts etc.  I make mistakes too, like 12 a day, just ask my

Again, Thanks for the generous thoughts and ideas, I will no doubt try them
all this weekend!

Chuck  WA9POU

-----Original Message-----
From: [] On
Sent: Tuesday, July 17, 2007 3:03 PM
Subject: Re: [Amps] Grid Dipping the Pi network in a new amp?


The antenna analyzer SHOULD have a source impedance of ~50 ohms.  If it
didn't it would cause a reflection (or possibly reduce or cancel a
reflection) if it presents an odd conjugate condition other than 50 ohms

Maybe I am not clear on what to do:  You're attaching a load impedance to
the input of the tank that emulates the desired tube operating load
impedance.  Chuck said that this is ~1800 ohms.  Given that a resistor of
that value is placed from that node (plate cap) to RF ground (along with the
shunt capacitance of the tube's output, Cp-g for grounded grid), and the
tank is properly adjusted, you should measure 50 ohms or a 1:1 SWR at the
output of the amplifier at desired frequency.  There is no other termination
at the output, only the source impedance of the antenna analyzer, 50 ohms.

You could go the other way, measure from the tank input and hang a 50 ohm
termination at the amp output, but most impedance bridges are too inaccurate
or won't resolve > 1K ohms.

This whole thing reminds me a bit of WG0M in Minneapolis a few years ago who
wanted me to design a matching network for his 160m "L" antenna.  I used
NEC2 to ballpark it.  He hooked it up and was mad because he still had 3:1
SWR.  I went over to help him, and it was immediately obvious what the
problem was:  The vertical portion of the "L" ran parallel and 3" away from
his 60' tower.  I never knew (nor did he tell me) he had a tower.  A little
fiddling and we had a match.

So if there's anything you're forgetting to tell me Chuck ...

73, Steve, K0SF

----- Original Message -----
From: "Bob Sullivan" <>
Sent: Tuesday, July 17, 2007 2:09:52 PM (GMT-0600) America/Chicago
Subject: Re: [Amps] Grid Dipping the Pi network in a new amp?


Been reading your very informative comments - thanks for taking the time to 

The idea of using an antenna analyzer is a great one but I thought about 
what you said below and was wondering if I should place a 52-ohm 
termination at the output of the tank before utilizing the analyzer (MFJ)?


73, Bob
Personal site:

At 08:22 AM 7/17/2007, you wrote:
>I just think it is something easy. For example, is your bandswitch 
>shorting the section of inductor that should be in play ?  Is your plate 
>choke much smaller than planned (that would have the effect of a pretty 
>strong inband zero) ?  Perhaps your plate blocking cap is too small a 
>value, or you're not setting the output T/R relay for transmit ?  It just 
>seems like too much work to take apart again.  These are all the things I 
>have had to straighten out with Bob, K0DD, but I assume you're way ahead 
>of him.
>Hey, believe me, we (and especially Bob) have made plenty of mistakes (I 
>can think of one at work last Thursday).  So, I guess, I would go check 
>the not-so-obvious obvious and make sure.
>If you use your antenna bridge, remember, that it IS the source impedance; 
>therefore, you're not putting a 50 ohm resistor at the tank output, the 
>antenna analyzer is sourcing 50 ohms and you're placing the 1800 ohm 
>resistor (if that is what you think the design plate impedance should be) 
>at the tube plate connection to ground.
>If you give me the tank component values and the config, I can quickly 
>calculate the input impedance.
>Another tip: when I do PI-L networks I usually set the middle node to the 
>conjugate mean of the input and output impedance.  Therefore, if the input 
>impedance is 1.8K (real), and the output is 50 ohms (real), in between the 
>two L networks should be: (1.8K*50)^.5 = 300 ohms real.  There may be Q 
>and efficiency benefits depending on the components used to use a 
>different intermediate impedance, but generally, the middle of the PI-L 
>should be at conjugate mean.
>Anyways, send me your data if you want some help.
>Steve, K0SF
>----- Original Message -----
>From: "Chuck Curran" <>
>Sent: Monday, July 16, 2007 7:00:04 PM (GMT-0600) America/Chicago
>Subject: RE: [Amps] Grid Dipping the Pi network in a new amp?
>To All:
>Thanks for the many valuable posts in regards to my dipping the Pi-Net
>output circuit.  I have read each and every one and learned a bit in the
>Steve, this last post of yours has caught my attention, since I just
>purchased a MFJ-259B about 6 weeks ago.  I have already tried most of the
>other suggestions, with no success.  The MFJ-259B was picked up for another
>issue, now resolved due to the ease of use of the antenna analyzer.
>I did not use any cleaners on the silver plated contact sections of the
>vacuum variables, but last time I checked, silver oxide was conductive.
>However, In order to resolve this issue by the weekend, I will dis-assemble
>all three key components, check everything again and then test -- I expect
>to discover my error during that endeavor!  I am considering completely
>removing the tune and load caps and the B&W 852 tank coil and then
>re-connecting all on the bench, and then try the analyzer.  I will do the
>fully assembled version first, since it is already there and ready.
>I am new to this list, and at least one of my posts seems to be missing,
>hope this one goes through OK.
>Thanks a million to all,
>-----Original Message-----
>From: [] On
>Sent: Monday, July 16, 2007 7:06 AM
>Subject: Re: [Amps] Grid Dipping the Pi network in a new amp?
>I'd dispense with the grid dip and get an antenna analyzer.  Terminate as I
>suggest (leave the tube in socket so that the tube parasitics are in place)
>and look at the output port with the analyzer.  Don't forget to key the
>output T/R relay (amp power off of course).
>This method works fine in HF and beyond.  I am doing 20 GHz work now at
>and using the same method.  The terminations are 0402 resistors, not leaded
>Also, not terminating the output will shift the result significantly, that
>is why I prefer the terminated method with a grid dip and I'll take the
>broader measurement.  But I don't use my grid dip any more.  Pay to get an
>analyzer, it is well worth it.
>Steve, K0SF
>----- Original Message -----
>From: "Tom W8JI" <>
>To: "STEVEN & NANCY FRAASCH" <>, "Chuck Curran"
>Sent: Sunday, July 15, 2007 4:27:14 PM (GMT-0600) America/Chicago
>Subject: Re: [Amps] Grid Dipping the Pi network in a new amp?
>The general condition where a grid dip meter dips the
>deepest is with the highest possible Q, NOT when the tank is
>loaded with a resistance.
>If you are not getting a dip and you are using a real grid
>dip meter, it is because you are looking at the tank
>incorrectly or have something wired wrong.
>Think of the tank as a transmission line. A normal
>pi-network tank when terminated acts like a line section
>with an electrical length of 130 degrees or so. In fact a
>tank perfectly  tuned on 7MHz for a 3000 ohm tube and 50 ohm
>load resonates when open circuited about 6.9 MHz, so you
>should see a dip near the operating frequency WITHOUT
>Of course this is still largely a useless test, sine it does
>not mean the tank is working as a matching network.
>The most useful test is to terminate the tube anode with a
>resistor that looks like the operating load impedance you
>want (use short leads to the resistor). Then you look at the
>output port and look for a low SWR.
>I'm going to side with Ian on this one. The largest dip is
>ALWAYS when there is no resistance, and the dip would be
>very close to the working frequency. You will get the
>poorest dip when you follow the advice to terminate the
>You have something else wrong, or have a bad dip meter.
>73 Tom
>Amps mailing list
>Amps mailing list

Amps mailing list

Amps mailing list

<Prev in Thread] Current Thread [Next in Thread>